Search schemes for random optimization algorithms that preserve the asymptotic distribution
Loading...
Files
Date
Authors
Advisor
Coadvisor
Graduate program
Undergraduate course
Journal Title
Journal ISSN
Volume Title
Publisher
Applied Probability Trust
Type
Article
Access right
Acesso restrito
Files
Abstract
Markovian algorithms for estimating the global maximum or minimum of real valued functions defined on some domain Omega subset of R-d are presented. Conditions on the search schemes that preserve the asymptotic distribution are derived. Global and local search schemes satisfying these conditions are analysed and shown to yield sharper confidence intervals when compared to the i.i.d. case.
Description
Keywords
random search algorithms, global optimization, search schemes, asymptotic distribution
Language
English
Citation
Journal of Applied Probability. Sheffield: Applied Probability Trust, v. 36, n. 3, p. 825-836, 1999.
